Kim Hyung-jun, CEO of Able Therapeutics, recently met with E-Daily’s pharmaceutical and biotech premium content service, Pharm E-Daily, at the company’s headquarters in Yeouido, Seoul. Using his laptop, he played back two audio files containing the same sentence one after the other. One was the original, and the other was a file from which a subtle tremor in the voice had been arbitrarily removed. Even after listening to them several times, it was difficult to detect a clear difference with the naked ear.
However, when the audio was converted into a frequency-based image on the screen, a difference between the two files became apparent. This demonstrates that artificial intelligence (AI) can distinguish speech characteristics—which are difficult for humans to detect—based on data.
“SPICK,” developed by Able Therapeutics, is a digital cognitive assessment platform that uses AI to analyze such speech characteristics—which are difficult for humans to distinguish—to evaluate the risk of cognitive decline. It was developed in response to the growing awareness that, as the aging population leads to a rapid increase in dementia patients, the importance of prevention and early screening is rising; however, existing tests alone have limitations in detecting mild cognitive impairment (MCI)—a condition associated with a relatively high risk of progressing to dementia—at an early stage.
According to the “Dementia Epidemiological and Status Survey” published by the Ministry of Health and Welfare and the National Dementia Center, three out of every ten seniors aged 65 and older have mild cognitive impairment (MCI)—considered a pre-dementia stage—and one in ten suffers from dementia. Able Therapeutics’ SPIC is a cognitive decline screening and diagnostic support tool that utilizes voice data to complement existing paper-based cognitive tests. It helps medical professionals integrate test results with medical history, previous cognitive test results, and other clinical information—such as imaging and blood test data—to determine the need for follow-up diagnostic testing.
SPIC received a Class 3 digital medical device manufacturing license from the Ministry of Food and Drug Safety last February. This marks the first such approval in South Korea for software medical devices designed to assess cognitive function. The company plans to fully launch its hospital-based service in October after completing regulatory and internal hospital procedures.
CEO Kim stated, “The company is utilizing approximately 12,500 voice files collected from healthy individuals, MCI patients, and dementia patients as training data,” adding, “When a user’s voice is inputted, the system calculates the risk of cognitive decline based on the speech patterns of the healthy and patient groups included in the training data.”
Why an IT Entrepreneur Entered the Cognitive Healthcare FieldCEO Kim did not start his career in the medical field. He is an entrepreneur who has spent over 25 years working in information technology (IT) and new business development. He began his career in the new business division of a large corporation and founded an IT company in 2013, personally commercializing various technology ventures such as enterprise messaging, system development, and beacons.
However, judging that the existing IT market alone had limitations for long-term growth, he began exploring the digital healthcare sector in 2017. In 2018, he also began studying AI at his alma mater’s graduate school to gain a deeper understanding of the field.
His decision to choose cognitive function assessment as his first business venture was driven not only by market potential but also by personal concerns. CEO Kim said, “While volunteering at a senior welfare center during my time as a company employee, I saw elderly people living alone with significantly impaired cognitive function, as well as families where an elderly spouse was caring for a dementia patient.” He added, “From that point on, I wondered if there was a way to detect early warning signs of cognitive decline even in environments lacking hospitals or specialized personnel.”
He continued, “Later, while reviewing research on Alzheimer’s disease, I noticed that changes such as slowed speech, hesitation, and difficulty finding words are reflected in the voice.” He added, “Since voice data can be collected using just a smartphone, I determined that it offers high accessibility and scalability and is a suitable digital biomarker for early screening.”
From the outset, he did not believe that technology alone could solve medical problems. Consequently, he conducted research with Professor Kim Woo-jun of the Department of Neurology at Seoul St. Mary’s Hospital, Professor Kim Jae-won of the Department of Child and Adolescent Psychiatry at Seoul National University Hospital, and Professor Emeritus Lim Jun-sik of Gachon University, and based on this work, he co-founded Able Therapeutics in 2021.
CEO Kim explained, “Medical professionals designed clinically meaningful test protocols and diagnostic criteria, the AI research team built the algorithms and data systems, and I was responsible for product development and commercialization.” He added, “Since digital medical devices are difficult to develop using just one of the following—medicine, AI, user experience, or commercialization—our multidisciplinary co-founding structure is the company’s most important foundation.”
Analyzing ‘Speaking Patterns’ Rather Than Correct Answers… Focus on MCI ScreeningSPEAK offers services tailored to various channels, including: △ a medical device version (11 questions, 15 minutes); △ a non-medical healthcare version (8 questions, 10 minutes); and △ a phone-based AI call version (4 questions, 5 minutes).
A key feature of the product is that it assesses the risk of cognitive decline by synthesizing vocal characteristics that are difficult for humans to detect through listening alone.
CEO Kim stated, “SPEAK does not score based on whether answers are correct. “Rather than focusing on ‘what is being said,’ it analyzes vocal patterns—such as speech rate, pauses, hesitations, and tremors—that reflect ‘how it is said,’” he said, adding, “Since the voice is converted into a mel-spectrogram image for the AI to identify patterns, this reduces variability caused by the examiner’s skill level and allows results to be quantified immediately.”
Based on these distinguishing features, in a confirmatory clinical trial conducted at eight university hospitals involving 399 participants—including a control group and patients with MCI and Alzheimer’s disease—SPEAK recorded a sensitivity of 79.6% and a specificity of 74.3% for MCI screening, with an area under the ROC curve (AUC) of 0.80, which indicates classification performance. The company explained that these results confirm the potential of SPIC as a voice-based screening tool that complements existing paper-and-pencil cognitive tests.
Of course, given that there were virtually no precedents for voice-based cognitive assessment software in Korea, the process of clinically validating Spik’s performance was not easy. The company had to start by aligning the intended use, target patient population, clinical criteria, and evaluation metrics with regulatory standards.
Launching
Hospital Services in October… Aiming for Multimodal and Overseas ExpansionAble Therapeutics is currently preparing for the non-reimbursed use of SPEAK in hospitals. CEO Kim stated, “We plan to launch hospital services around October,” adding, “Initially, rather than rapidly increasing the number of hospitals, we are prioritizing verifying the test administration rate and repeat usage rate within actual clinical workflows and accumulating real-world evidence (RWE) that medical staff can trust.”
Outside of medical institutions, the company is also commercializing a non-medical healthcare version. The tablet-based healthcare version has been supplied to daycare centers and home-based senior care centers since 2024 and is being used for the daily cognitive health management of the elderly.
The AI Call version, which involves performing simple speaking tasks over the phone, has been applied since last year to a cognitive health management service targeting medically underserved elderly residents in Bucheon City. CEO Kim said, “While the hospital-grade medical device supports the screening of high-risk groups for cognitive decline and assists medical professionals’ decision-making, the healthcare version serves to expand access points so that older adults can continuously monitor their own cognitive health even outside of hospitals.”
Able Therapeutics is targeting approximately 400 million won in revenue this year, based on the full-scale commercialization of the hospital-grade SPIC and the expansion of its core healthcare business.
CEO Kim also outlined his goal to grow Able Therapeutics into a “cognitive healthcare platform” rather than simply a voice-based testing company. His mid- to long-term objectives include △enhancing algorithm performance, △expanding the product pipeline, and △developing international versions.
Able Therapeutics is currently working on developing international versions of the product. First, the company collaborated with local medical professionals in Taiwan to develop tasks in Standard Chinese and completed a Phase 1 clinical trial involving 331 participants. CEO Kim stated, “Overseas localization is not simply a matter of translating Korean content and exporting it as-is.” He added, “While the core acoustic analysis structure can be shared, memory and language tasks are influenced by cultural and educational backgrounds. Therefore, we must redesign the task sets in collaboration with local experts and validate performance by securing data from both healthy controls and patients.”
